Reading a proxy history

A testing proxy shows the request and the response for an application you are allowed to test. The reading order is who is signed in, which field the client sent, and which control the response is missing.

Lesson 25 of 26 · 14 min read · practice only with authorization

Diagram

From a captured exchange to a missing control

Follow the path. One example at a time.

Step 1 of 4

Scope

The proxy is pointed at a lab or an application you were authorized to test. Other traffic is not the lesson.

Illustration: the scope comes first
# You are looking at an application you are allowed to test.
# The proxy is a window onto that application.
What changes the outcome
Note

A long export of every request.

Claim

The proxy “found something.”

Reader

Cannot see the missing control.

The example on the open step is one illustration, or the fix you can paste into an application you own.

What this page is made of

4 pieces. Each one says what fails, who is affected, what you would notice, and what you change.

  1. The proxy is a window

    What fails
    It does not create the bug. It shows bytes. A history full of requests is not a finding until you can point at one exchange.
    Who is affected
    Users of the application, if the exchange shows a real missing control.
    What you would notice
    A report that attaches the history and does not name a control.
    What you change
    Pick one request. Write the four fields. Stop.
  2. Who is signed in

    What fails
    The cookie or the authorization header is the subject. If it is empty, say so. If it is a lab user, name the lab user. Do not paste a production token.
    Who is affected
    The account that subject can act as.
    What you would notice
    A screenshot that includes a live authorization header.
    What you change
    Redact tokens. Describe the subject: “Avery’s lab session.”
  3. The client field

    What fails
    Query parameters, JSON fields, and headers the client chose are the inputs. The question is whether the server treated them as data, as a URL to fetch, or as an object id.
    Who is affected
    Whoever is affected by that treatment, from the matching lesson.
    What you would notice
    A note that says “parameter” and does not say which one.
    What you change
    Name the field. Name the control that should bind, encode, or authorize it.
  4. What you do not automate here

    What fails
    Repeating the same request with a long list of inputs is not required to recognize a missing control, and this course does not teach that automation.
    Who is affected
    Anyone whose account would be hit by unsolicited traffic.
    What you would notice
    A write-up that is mostly tool configuration.
    What you change
    Stay with the single exchange and the control. If the scope forbids a test, the proxy stays closed.

A proxy records. It does not decide.

Testing proxies sit between a browser and an application so you can read the HTTP you would otherwise only see as a screen. That is the whole job. The bug, if there is one, is in the application: a missing ownership check, a missing token, text treated as markup, a cookie with no flags. The proxy is how you noticed. The finding has to be readable by someone who will not open the proxy.

Point it at a lab or at a target your scope names. Your ordinary browsing does not belong in the same browser profile. A proxy that sees your mail and your bank is a copy of your mail and your bank. Use a separate profile. Close it when the lab is done.

Read one exchange in order

Method and path say what was attempted. The cookie or authorization header says who the server thought it was. Redact the secret and write the lab user’s name. The client field is the query parameter, the JSON property, or the header the browser added because of something you control in the lab. The response says whether the server encoded, authorized, or refused. Security headers and set-cookie flags are part of the response, not a separate hobby.

If you cannot fill those lines, you do not have a finding yet. You have a transcript. Transcripts are how reports get ignored. One exchange and a named control from this course is a finding. Automation that throws many inputs at the field is not required for that sentence, and it is out of scope for this page.

Authorization still applies

A proxy makes it easy to send a request the user interface did not offer. That does not expand the scope. You may read an exchange on an application you were allowed to test. You may not point the same profile at a host that was not named, and you may not keep a production token because it was “already in the history.” Delete it. If it was real, rotate it.

You would notice a note that says “the proxy found XSS” with no field and no encoding fix, or a screenshot with a live cookie. The people affected are the users of the real control you failed to name. Rewrite the note before you file it.

What you change

Change the application: encode, authorize, require the token, set the cookie flags. The proxy configuration does not ship. In the report, keep the five lines from the reading order and the lesson you are pointing at. Link the evidence lab if the reader needs practice on a trace that is already collected, which is the same muscle without a live proxy.

When the vendor’s proxy and this page use different words for the history view, use theirs for the click and this page for the sentence. The sentence is the part that survives a new version of the tool.

Check yourself

Recognition and the fix. Nothing here is a procedure.

1 / 3

A proxy history shows a lab user’s request for an invoice id, and the response body belongs to a different lab user. What do you write down?

Sibling techniques

Pages at the same depth, plus a lab and a glossary card for this one.

  1. MapA method for any tool you have not met yetNew tools show up every year.
  1. 01Diff a scan against the inventoryA scanner produces observations and severities.

Practice

  1. LabControl benchA failure is described. Choose the control that changes the outcome.